Re: [platform-releng-dev] broken metadata? M-build respin scheduled for

[David M Williams <david_williams@xxxxxxxxxx>]

Well, it looks like as soon as I hit the
"send" button, the build did in fact finish, and a quick scan
of it's repo does have the correct versions of EF and httpcomponents in
it. 

So ... guess I'll assume it is a fine
build, unless/until someone reports a problem. 

http://download.eclipse.org/eclipse/downloads/drops4/M20141205-1400/

(You can even see the first version
of our download page using the Solstice theme ... and, not to worry, I've
already made a few improvements since that version, and will be making
more before Monday. )

Weird stuff happens, I guess. 




From:      
 David M Williams/Raleigh/IBM@IBMUS
To:      
 "Eclipse platform
release engineering list." <platform-releng-dev@xxxxxxxxxxx>,

Date:      
 12/06/2014 12:19 AM
Subject:    
   Re: [platform-releng-dev]
broken metadata? M-build respin scheduled for 2 PM -- status
Sent by:    
   platform-releng-dev-bounces@xxxxxxxxxxx

---

Believe it, or not ... it *is* still
running. But, not sure I'd trust it ... now after 10 hours of running (at
least) what is normally a 3 to 4 hour job. It would probably be fine if
it ever finished, but ... something is amiss on the build machine. Even
the N-build, that started at 8 PM, is *still* in its very initial phase
(of clean up and checkout of source) .... something that usually takes
the "long" time of 20 or 30 minutes (now gong on for 4 hours).

Unless something changes in the next 20 to 30 minutes, I am going to cancel
both of them, and try again tomorrow, after making sure "everything"
is cleaned up. After the initial failure today, related to NFS, things
might have been left in a bad state? Or .. perhaps There are still NFS
problems? Normally we do allocate 2.5G "max heap", but, for some
reason there appears to be 6G of memory swapped to disk for one of the
processes. I know from desktop use, at least, once something gets in a
"bad swap state", it can not really catch up, just spends all
its time swapping memory pages for one bit of code or another. (And, by
the way, pretty sure "it is not us" ... the build still runs
normally on my local test system, with "signing" being the only
substantial difference. 

Hope for the best ... 

Apologies for the delay, 
  





From:        David
M Williams/Raleigh/IBM@IBMUS 
To:        "Eclipse
platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date:        12/05/2014
01:56 PM 
Subject:        Re:
[platform-releng-dev] broken metadata? M-build respin scheduled for 2 PM
Eastern, today, Friday 12/5 
Sent by:        platform-releng-dev-bounces@xxxxxxxxxxx

---

The first re-build attempt failed, with what seems like some sort of transient
"NFS file system issue" (Bug
454272 )
so I have rescheduled another for 2 PM to try again. 





From:        David
M Williams/Raleigh/IBM@IBMUS 
To:        "Eclipse
platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date:        12/05/2014
08:51 AM 
Subject:        Re:
[platform-releng-dev] broken metadata? M-build respin scheduled for 10
AM Eastern, today, Friday 12/5 
Sent by:        platform-releng-dev-bounces@xxxxxxxxxxx

---

Starting at 10, the build will be done approximately 1 PM, with Unit test
following much later (i.e. after most have started their weekend :)

Thanks for finding and raising the issue, and providing a humbling reminder
that every build must be "good". 

Again, sorry for the churn. 




From:        Andreas
Sewe <andreas.sewe@xxxxxxxxxxxxxx> 
To:        David
M Williams/Raleigh/IBM@IBMUS, 
Cc:        platform-releng-dev@xxxxxxxxxxx
Date:        12/05/2014
05:08 AM 
Subject:        Re:
[platform-releng-dev] Yesterday's 4.4 M-build: broken metadata?
Sent by:        platform-releng-dev-bounces@xxxxxxxxxxx

---

Hi David,

> Well, by "idea" ... I can confirm it is broken. I fixed
our pom's last
> night (with no new bug, since it was an incomplete fix for bugs 453677
> and 453919 -- entirely my fault).
> 
> I was not aware of anyone making use of the M-build repositories this
> early, so, if you do not want to wait next Wednesday, it would be
> entirely reasonable for you to ask for a respin.

Well, as I'm lazy and don't want to update the Code Recommenders target
platform to a new M-build update site each time one is announced, I
simply used the M-builds composite site; apparently, this exposes me to
new M-builds earlier than the less lazy folks.

> Apologies for the break, and especially not properly announcing it.

No problem. Detecting issues early is much better than detecting issues
late in the game, so I think overall things "worked" (as in "broke")
as
expected.

That being said, it would be great if your could do a re-spin, yes.

(The "quick fix you described does work, but if doing a re-spin is
not
too much trouble, I'd rather have you do that instead of committing and
reverting a quick fix for a day or two.)

Best wishes,

Andreas

-- 
Codetrails GmbH
The knowledge transfer company

Robert-Bosch-Str. 7, 64293 Darmstadt
Phone: +49-6151-276-7092
Mobile: +49-170-811-3791
http://www.codetrails.com/

Managing Director: Dr. Marcel Bruch
Handelsregister: Darmstadt HRB 91940
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
from this list, visit
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev

_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
from this list, visit
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
from this list, visit
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe
from this list, visit
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev